In general, in order to ensure safe driving, unmanned autonomous vehicles detect an obstacle positioned in a driving direction and determine a driving direction by using the detected information. However, when an unmanned autonomous vehicle needs to carry out a particular purpose, while driving outdoors, rather than indoors, it is important for the vehicle not to collide with an obstacle present on the ground.
Running on the roadway, such an unmanned autonomous vehicle should drive, without colliding with the sidewalk protruded to be higher than the road, and thus, it is necessary for the unmanned autonomous vehicle to detect a curb as a boundary stone between the roadway and the sidewalk. A curb is an important mark present on the road, and thus, finding a curb ensures safe driving of the unmanned autonomous vehicle.
In a related art curb detecting method, a curb is detected by utilizing sensor information obtained mainly from the side of a curb. Namely, in the related art, since the side of a curb is perpendicular to the ground, a curb is detected by using the characteristics that distance sensor information is rapidly changed in a curb portion.
However, an amount of the sensor information obtained from the side of a curb is so small that, if only the side information of the curb is used, curb detection performance is easily affected by an external factor, noise of a sensor, and the like, and it is difficult to recognize a curb present at a distance from a vehicle. For example, in public facilities, the side of a curb is short with a length ranging from 6 cm to 15 cm, and thus, only a small amount of information is obtained from the side of a curb. Thus, when a distance to a curb is long, it may be difficult to obtain information sufficient for differentiating the ground and a curb. Namely, it is difficult to detect a curb at a distance with the related art method.
In addition, in the related art method for detecting a curb, the side of a curb which is partially damaged or which is partially covered with fallen leaves cannot be accurately detected and sensor information includes a great amount of errors, resulting in failure to accurately detect a curb.
As a solution to the problem, Korean Patent Laid Open Publication No. 10-2012-0052691 (published on May 24, 2012) (hereinafter, referred to as a “related art”) discloses a method for detecting a curb as a boundary stone between a roadway and a sidewalk by comparing information obtained by a distance sensor installed in a vehicle and a previously designed curb model.
According to the related art, a curb can be detected stably under various environmental conditions by using upper and lower horizontal planes of a curb, as well as the side of a curb, and mathematically comparing a curb model with distance sensor information.
However, the related art has a problem in that only a predetermined region from a vehicle can be detected, only a following space in which a vehicle has moved in a section sensed by a distance sensor can be recognized, and a running vehicle cannot recognize an available driving space because it fails to detect a curb in a large space.
In addition, a recognition rate may be lowered in a section in which a height of a curb is low, and several sections of a curb cannot be simultaneously measured in real time.